Yougang Lyu
@yougang_lyu
Followers
75
Following
288
Statuses
32
PhD student @irlab_amsterdam & @UvA_Amsterdam | Intern @Baidu_Inc | Working on aligning language agents with human
Joined October 2021
Thank you for having me!
Today, Yougang Lyu @yougang_lyu from the University of Amsterdam is giving an #IRTalk entitled "Aligning language models with human understanding and behavior". Details: @GlasgowCS
@ir_glasgow
0
0
1
0
0
0
RT @ZihanWa54274484: Thrilled to share that our paper, "A Cooperative Multi-Agent Framework for Zero-Shot Named Entity Recognition," has beā¦
0
2
0
RT @hunman_centerRS: Call for Papers: Workshop on Human-Centered Recommender Systems @TheWebConf 2025 Updated Submission Deadline: Januaryā¦
0
3
0
RT @SIGIRConf: š£ š£ #sigir2025 call for workshops Let's brainstorm on IR, RecSys, GenAI, RAG, and more. - Submissiā¦
0
14
0
RT @furongh: I know there are already plenty of tips out there on how to write an effective rebuttal, but I thought Iād share mine as well.ā¦
0
17
0
RT @iclr_conf: We are extending the ICLR discussion phase by 6 days. The new timeline is: Nov 26th: Last day the PDF may be revised (no chā¦
0
16
0
RT @hunman_centerRS: Call for Papers: Workshop on Human-Centered Recommender Systems at TheWebConf 2025 We invite you to submit your reseaā¦
0
4
0
š§How to fine-tune LLMs to solve complex knowledge-aware QA tasks? Our #EMNLP2024 Main paper "KnowTuning: Knowledge-aware Fine-tuning for Large Language Models" proposes KnowTuning to improve the fine-grained and coarse-grained knowledge awareness of LLMs for solving complex knowledge-intensive tasks. āļøCome check out our poster on Nov 12 (Tue) at 16:00.
0
2
4
@theryanliu Great work, congrats! We also found a similar phenomenon in our recent experiments, using COT on high-accuracy (using standard prompt) test data rather reduces performance. It looks like using COT on these data introduces additional distracting noise.
0
0
2
RT @SIGIRConf: š£ š£ #sigir2025 call for full papers Share your research in IR, RecSys, GenAI, RAG, and more. - Absā¦
0
23
0
@ZhengxiangShi Thanks for your explanation, itās good to see some work explaining dpo-like losses, because only using dpo seems will break the output format. Congrats again!
0
0
1
RT @zhaochun_ren: Thrilled to have won the Best Paper Award at the top conference on recommender systems, #RecSys2024 . Congratulations toā¦
0
5
0